INTELLIGENT IDENTIFICATION METHOD OF AQUIFER DISTRIBUTION IN COMPLEX EDGE-WATER OIL AND GAS RESERVOIRS

Abstract

An intelligent identification method of aquifer distribution in complex edge-water oil and gas reservoirs is provided. The problem, that targeted water control countermeasures cannot be proposed due to unclear identification of edge-water distribution in complex edge-water reservoirs, is resolved. Based on geological data and production data of a single well, establish a numerical simulation model of a water influx unit for simulating inflow dynamic of complex edge-water influx; by a genetic algorithm, correct characteristic parameters of an aquifer unit, including volume and water influx of the aquifer unit; automatically fit dynamic production data calculated by the model with actual dynamic production data, to obtain optimal characteristic parameters of the aquifer unit; and assign the characteristic parameters to the aquifer unit for inversion to determine aquifer distribution. The method has simple steps and accuracy by comparing the result of aquifer distribution inversion and that of numerical simulator.

Description

TECHNICAL FIELD

The present disclosure relates to an intelligent identification method of aquifer distribution in complex edge-water oil reservoirs, and belongs to the technical field of oil and gas field development.

BACKGROUND ART

For an oil reservoir with edge water, the oil reservoir will be flooded due to the influence of water influx in a development stage of the oil reservoir. The overall production regime of a production well will be affected after water breakthrough, thereby affecting the ultimate recovery factor of the oil reservoir. Therefore, it is of great guiding significance to accurately understand the aquifer distribution characteristics of an oil reservoir for selecting a reasonable water control measure for the oil reservoir and adjusting a development plan. In the studies of aquifer characteristics, aquifer characteristics are characterized by means of characteristic parameters such as the aquifer volume and the water influx. However, these studies cannot characterize the distribution of aquifers, so the requirements of targeted treatment and adjustment of oil reservoirs with water influx cannot be met.

At present, the distribution location of aquifers can be characterized by a physical model establishing method and a numerical simulation method. However, the two methods still have certain limitations. For the physical model establishing method, the water influx process of an oil reservoir is a large-scale fluid movement, and there are insurmountable problems in realization of a physical model similar to a prototype. With development of computer technology, the numerical simulation method has a wider application scope, higher calculation efficiency, and lower cost. Many scholars use the numerical simulation method to simulate a dynamic process of water influx. At present, the numerical simulation method simulates the dynamic process of water influx mostly by establishing an aquifer based on geological data, and verifies accuracy of the model through history matching. There are two problems in this process: (1) for a numerical model with measurement errors or lack of geological data, the result of the established water model is not convincing enough; and (2) in the process of history matching of the model, there are too many parameters to be adjusted, so that the accuracy of the established aquifer model cannot be fully explained.

In general, the current methods for characterizing distribution of aquifers in complex edge-water oil reservoirs have certain limitations. Therefore, an intelligent identification method of aquifer distribution suitable for complex edge-water oil reservoirs is urgently needed.

SUMMARY

The objective of the present disclosure is: to solve the problem that targeted water control countermeasures cannot be carried out in the process of water influx of oil reservoirs due to unclear understanding of aquifer distribution in complex edge-water reservoirs, thereby affecting the ultimate recovery factor of the oil reservoirs, and based on single-well geological data and production data, aquifer characteristics are inverted on the basis of a numerical simulation method and an intelligent optimization algorithm, and re-identification of the aquifer distribution is realized.

To achieve the above objective, the present disclosure provides an intelligent identification method of aquifer distribution in complex edge-water oil reservoirs, including:

• first, establishing a numerical simulation model of a water influx unit: based on an aquifer unit, a production well unit and an infill unit, calculating the pressure, saturation and water cut of each unit by a material balance equation and a waterflood front advancing equation, and establishing the numerical simulation model of the water influx unit;
• second, automatically fitting aquifer characteristic parameters: based on the numerical simulation model of the water influx unit, with actual dynamic production data as a reference, automatically correcting aquifer unit volume and aquifer unit cumulative water influx by a genetic algorithm to make data predicted by the model match the actual dynamic production data, and using the corrected aquifer unit volume and corrected aquifer unit cumulative water influx as aquifer characteristic parameters after the model is automatically fitted; and
• third, inverting aquifer distribution: after obtaining optimal solutions of the aquifer characteristic parameters, assigning the aquifer unit volume and the aquifer unit cumulative water influx to a discretized aquifer unit, establishing an aquifer unit bar chart, and determining aquifer distribution by the bar height of each unit.

In the intelligent identification method of aquifer distribution in complex edge-water oil reservoirs, the step of establishing the numerical simulation model of the water influx unit specifically includes:

• first, simply characterizing an edge-water oil reservoir: discretizing an edge aquifer into the aquifer unit communicated with the production well unit; discretizing a position with clear geological knowledge into the infill unit; establishing a simple edge-water oil and gas reservoir unit system based on the aquifer unit, the production well unit and the infill unit; and characterizing the aquifer unit by the aquifer unit volume and the aquifer unit cumulative water influx;
• second, based on the aquifer unit, the production well unit and the infill unit, calculating the pressure of each unit by the material balance equation, which is expressed as:
• ${\displaystyle \sum _{j=1}^{n_{\text{well}}}{T}_{ji}^t\left(p_j^t-p_i^t\right)+w_i^t=c_t\cdot V_i^t\cdot \frac{d{p}_i^t}{dt},}$
• where
• $T_{ij}^t$
• is an average conductivity between the aquifer unit i and the production well unit j at time t, m³/(d·MPa); n_well is a quantity of the production well units and infill units, dimensionless;
• $p_i^t$
• is a aquifer pressure of the aquifer unit i at time t, MPa;
• $p_j^t$
• is average bottom hole pressure of the production well unit j at time t, MPa;
• $w_i^t$
• is the cumulative water influx of the aquifer unit i at time t, m³/d; c_t is a comprehensive formation compressibility, MPa^-1; and
• $V_j^t$
• is the aquifer volume of the aquifer unit i at time t, m³;
• third, using a finite difference method to solve a pressure solution of the material balance equation advancing from an old time step t to a new time step t+1, with a constructed pressure matrix as follows:
• $\begin{array}{c}\left(\begin{array}{c}{p}_1^t\\ p_2^t\\ ⋮\\ p_n^t\end{array}\right)=\left(\begin{array}{c}{p}_1^{t+1}\\ p_2^{t+1}\\ ⋮\\ p_n^{t+1}\end{array}\right)\cdot \left(\begin{array}{cccc}\frac{\Delta t}{c_t\cdot V_1^{t+1}}{\displaystyle \sum _{a=1}^n{T}_{1a}^t+1}& -\frac{\Delta t}{c_t\cdot V_1^{t+1}}\cdot T_{12}^t& \cdots & -\frac{\Delta t}{c_t\cdot V_1^{t+1}}\cdot T_{1n}^t\\ -\frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot T_{21}^t& \frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot {\displaystyle \sum _{a=1}^n{T}_{2a}^t+1}& \cdots & -\frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot T_{2n}^t\\ ⋮& ⋮& \ddots & ⋮\\ -\frac{\Delta t}{c_t\cdot V_n^{t+1}}\cdot T_{n1}^t& -\frac{\Delta t}{c_t\cdot V_n^{t+1}}\cdot T_{n2}^t& \cdots & \frac{\Delta t}{c_t\cdot V_n^{t+1}}\cdot {\displaystyle \sum _{a=1}^n{T}_{na}^t+1}\end{array}\right)-\\ \left(\begin{array}{c}\frac{\Delta t}{c_t\cdot V_1^{t+1}}\cdot Q_1^t\\ \frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot Q_2^t\\ ⋮\\ \frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot Q_n^t\end{array}\right),\end{array}$
• where α is other units when a unit n is a research object, including the aquifer unit i and the production well unit j, dimensionless;
• $Q_n^t$
• represents flow of the unit n at time t, and since the unit n includes the aquifer unit i and the production well unit j, when
• $Q_n^t$
• is positive, it is equal to the cumulative water influx, and when it is negative, it is equal to the production well output, m³/d; and Δt represents the time step, d; and
• fourth, according to the calculated pressure of each unit, calculating the saturation and water cut of each unit by using the waterflood front advancing equation, which is:
• ${\left(\frac{dx}{dt}\right)}_{Sw}=\frac{Q}{A\varphi }\cdot \frac{\partial f_w}{\partial S_w},$
• where x is a position of a water influx channel, m; Q is total flow of fluid, m³; S_w is the water saturation at position x, dimensionless; ƒ_w is partial flow of a water phase, dimensionless; A is a cross-sectional area of a seepage, m²; and ϕ is porosity, dimensionless.

In the intelligent identification method of aquifer distribution in complex edge-water oil reservoirs, the step of automatically fitting the aquifer characteristic parameters specifically includes:

• first, establishing characteristic parameter vectors and an objective function of the aquifer unit according to the established numerical simulation model of the water influx unit, the established characteristic parameter vectors including the aquifer unit volume and the aquifer unit cumulative water influx, with an expression as:
• $m=⌊V_1,\mathrm{...},V_i,\mathrm{...},V_{n_{water}},w_1,\mathrm{...},w_i,\mathrm{...},w_{n_{water}}⌋,$
• and the established objective function is expressed as:
• $Y{\left(m\right)}_{\min }=\exp \frac{\text{sum}{\left[\text{ydata}-F(m)\right]}^2}{n_{water}},$
• where m is the characteristic parameter vector of the aquifer unit; Y(m) is the objective function of the aquifer unit; ydata is observed dynamic data; and F(m) is dynamic data calculated by the model correcting the characteristic parameters of the aquifer unit;
• second, setting constraints on the aquifer unit volume, and considering that the aquifer unit volume is equal to a size of the entire edge-water aquifer; and
• third, automatically fitting the aquifer unit volume and the aquifer unit cumulative water influx by the genetic algorithm, until the error value between the data predicted by the model and the actual dynamic production data is minimum, thereby obtaining the corrected aquifer unit volume and the corrected aquifer unit cumulative water influx.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a diagram showing a technical route of the present method.

FIG. 2 is a schematic diagram showing aquifer distribution inversion.

FIG. 3 is a curve graph showing relative permeability for an established numerical simulation.

FIG. 4 is a schematic diagram showing concave aquifer distribution.

FIG. 5a is a diagram showing a fitting curve of the water cut of a well P1.

FIG. 5b is a diagram showing a fitting curve of the water cut of a well P2.

FIG. 5c is a diagram showing a fitting curve of the water cut of a well P3.

FIG. 5d is a diagram showing a fitting curve of the water cut of a well P4.

FIG. 5e is a diagram showing a fitting curve of the water cut of a well P5.

FIG. 5f is a diagram showing a fitting curve of the water cut of a well P6.

FIG. 6 is a diagram showing the inversion result of aquifer distribution automatically fitted by a model.

FIG. 7 is a diagram showing the water saturation distribution calculated by the model.

DETAILED DESCRIPTION OF THE EMBODIMENTS

The present disclosure is further described below with reference to the embodiment and accompanying drawings.

The present disclosure provides an intelligent identification method of aquifer distribution in complex edge-water oil reservoirs, FIG. 1 is a diagram showing a technical route of the present disclosure, and the method includes the following steps:

First, a numerical simulation model of a water influx unit is established: based on an aquifer unit, a production well unit, and an infill unit, the pressure, saturation and water cut of each unit are calculated by a material balance equation and a waterflood front advancing equation to establish the numerical simulation model of the water influx unit.

Second, aquifer characteristic parameters are automatically fitted: based on the numerical simulation model of the water influx unit, with actual dynamic production data as a reference, two characteristic parameters, namely volume and water influx of the aquifer unit are automatically corrected by a genetic algorithm to make data predicted by the model match the actual dynamic production data, and the corrected volume and water influx of the aquifer unit are used as aquifer characteristic parameters after the model is automatically fitted.

Third, aquifer distribution is inverted: after optimal solutions of the aquifer characteristic parameters are obtained, the two characteristic parameters, namely the volume and water influx of the aquifer unit are assigned to a discretized aquifer unit, an aquifer unit bar chart is established, and aquifer distribution is determined by the bar height of each unit. The final aquifer distribution inversion result are shown in FIG. 2.

Further, the step of establishing the numerical simulation model of the water influx unit specifically includes the following steps:

First, an edge-water oil reservoir is simply characterized: an edge aquifer is discretized into the aquifer unit communicated with the production well unit; a position with clear geological knowledge is discretized into the infill unit; a simple edge-water oil reservoir unit system based on the aquifer unit, the production well unit and the infill unit is established; and the aquifer unit is characterized by two parameters, namely the volume and the cumulative water influx of the aquifer unit.

Second, based on the aquifer unit, the production well unit and the infill unit, the pressure of each unit is calculated by the material balance equation, which is expressed as:

${\displaystyle \sum _{j=1}^{n_{\text{well}}}{T}_{ji}^t\left(p_j^t-p_i^t\right)+w_i^t=c_t\cdot V_i^t\cdot \frac{d{p}_i^t}{dt},}$

where

$T_{ij}^t$

is an average conductivity between the aquifer unit i and the production well unit j at time t, m³/(d·MPa); n_well is the number of the production well units and infill units, dimensionless;

$p_i^t$

is aquifer pressure of the aquifer unit i at time t, MPa;

$p_j^t$

is average bottom hole pressure of the production well unit j at time t, MPa;

$w_i^t$

is the water influx of the aquifer unit i at time t, m³/d; c_t is a comprehensive formation compressibility, MPa^-1; and

$V_j^t$

is the aquifer volume of the aquifer unit i at time t, m³.

Third, a finite difference method is used to solve a pressure solution of the material balance equation advancing from an old time step t to a new time step t+1, with a constructed pressure matrix as follows:

$\begin{array}{l}\left(\begin{array}{c}{p}_1^t\\ p_2^t\\ ⋮\\ p_n^t\end{array}\right)=\left(\begin{array}{c}{p}_1^{t+1}\\ p_2^{t+1}\\ ⋮\\ p_n^{t+1}\end{array}\right)\\ \cdot \left(\begin{array}{cccc}\frac{\Delta t}{c_t\cdot V_1^{t+1}}\cdot {\displaystyle \sum _{a=1}^n{T}_{1a}^t+1}& -\frac{\Delta t}{c_t\cdot V_1^{t+1}}\cdot T_{12}^t& \cdots & -\frac{\Delta t}{c_t\cdot V_1^{t+1}}\cdot T_{1n}^t\\ -\frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot T_{21}^t& \frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot {\displaystyle \sum _{a=1}^n{T}_{2a}^t+1}& \cdots & -\frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot T_{2n}^t\\ ⋮& ⋮& \ddots & ⋮\\ -\frac{\Delta t}{c_t\cdot V_n^{t+1}}\cdot T_{n1}^t& -\frac{\Delta t}{c_t\cdot V_n^{t+1}}\cdot T_{n2}^t& \cdots & \frac{\Delta t}{c_t\cdot V_n^{t+1}}\cdot {\displaystyle \sum _{a=1}^n{T}_{na}^t+1}\end{array}\right)\\ -\left(\begin{array}{c}\frac{\Delta t}{c_t\cdot V_1^{t+1}}\cdot Q_1^t\\ \frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot Q_2^t\\ ⋮\\ \frac{\Delta t}{c_t\cdot V_2^{t+1}}\cdot Q_n^t\end{array}\right),\end{array}$

where α is other units when a unit n is a research object, including the aquifer unit i and the production well unit j, dimensionless;

$Q_n^t$

represents flow of the unit n at time t, and since the unit n includes the aquifer unit i and the production well unit j, when

$Q_n^t$

is positive, it is equal to the water influx, and when it is negative, it is equal to the production well output, m³/d; and Δt represents the time step, d.

Fourth, according to the calculated pressure of each unit, the saturation and water cut of each unit are calculated by using the waterflood front advancing equation, which is:

${\left(\frac{dx}{dt}\right)}_{Sw}=\frac{Q}{A\varphi }\cdot \frac{\partial f_w}{\partial S_w},$

where x is a position of a water influx channel, m; Q is total flow of fluid, m³; S_w is the water saturation at position x, dimensionless; ƒ_w is partial flow of a water phase, dimensionless; A is a cross-sectional area of a seepage, m²; and ϕ is porosity, dimensionless.

In the intelligent identification method of aquifer distribution in the complex edge-water oil reservoirs, the step of automatically fitting the aquifer characteristic parameters specifically includes the following steps:

First, characteristic parameter vectors and an objective function of the aquifer unit are established according to the established numerical simulation model of the water influx unit, and the established characteristic parameter vectors include the volume and water influx of the aquifer unit, with an expression as:

$m=⌊V_1,\mathrm{...},V_i,\mathrm{...},V_{n_{water}},w_1,\mathrm{...},w_i,\mathrm{...},w_{n_{water}}⌋,$

and the established objective function is expressed as:

$Y{\left(m\right)}_{\min }=\text{exp}\frac{\text{sum}{\left[\text{ydata}-F\left(m\right)\right]}^2}{n_{water}},$

where m is the characteristic parameter vector of the aquifer unit; Y(m) is the objective function of the aquifer unit; ydata is observed dynamic data; and F(m) is dynamic data calculated by the model correcting the characteristic parameters of the aquifer unit.

Second, constraints are set on the volume of the aquifer unit, where the volume of the aquifer unit is considered as equal to a size of the entire edge-water aquifer.

Third, the two characteristic parameters, namely the volume and the water influx of the aquifer unit are automatically fitted by the genetic algorithm, until the error value between the data predicted by the model and the actual dynamic production data is minimum, thereby obtaining two aquifer characteristic parameters, namely the corrected volume and water influx of the aquifer unit.

The model established by the method of the present disclosure is applied to a specific example as follows:

Using the model established herein, the calculation results of dynamic characteristics of water influx were compared with the results obtained by the numerical simulator. The numerical simulation model was verified by a black-oil model. The model grid was set to 50×50×1, the horizontal step size of each grid block was set to 10 m, the longitudinal step size was set to 20 m, and the simulation duration was 2000 days. The density of crude oil was 776 kg/m³, the viscosity of crude oil was 20 cp, and the compressibility of crude oil was 0.005 MPa^-1. The relative permeability curve of an oil reservoir was set as shown in FIG. 3. In the numerical simulation model, the numerical edge-water aquifer size was set to 1.8×10⁷ m³, and the aquifer distribution type is concave aquifer distribution, as shown in FIG. 4. Production wells were set as a constant fluid volume of production, respectively P1: 100 m³/d, P2: 50 m³/d, P3: 100 m³/d, P4: 200 m³/d, P5: 100 m³/d, and P6: 200 m³/d, and simulated production was 2000 d.

Production data of 2000 d was extracted. The water cut of the production wells P1-P6 predicted by the numerical simulator was used as an objective function of the model, and history matching of two characteristic parameters, namely the volume and water influx of the aquifer unit was carried out. It can be seen that in the validation of the concave aquifer model, the fitting effect of the water cut curve was good, the water breakthrough time of the production wells was consistent with the final water cut, and the fitting results of the water cut were shown in FIGS. 5a-5f.

According to the fitting results, the characteristic values of the volume and the water influx of a concave aquifer unit were obtained. These characteristic parameters were combined together, and the aquifer distribution of concave edge water could be obtained intuitively, as shown in FIG. 6. On the basis of the characteristic parameters, namely the fitted volume and water influx of the aquifer unit, the dynamic characteristics of water influx were obtained by model inversion, as shown in FIG. 7.

Claims

1. An intelligent identification method of aquifer distribution in complex edge-water oil and gas reservoirs, comprising: S100, establishing a numerical simulation model of a water influx unit: based on an aquifer unit, a production well unit and an infill unit, calculating pressure, saturation and water cut of each unit by a material balance equation and a waterflood front advancing equation to establish the numerical simulation model of the water influx unit;S200, automatically fitting aquifer characteristic parameters: based on the numerical simulation model of the water influx unit, with actual dynamic production data as a reference, automatically correcting volume and cumulative water influx of the aquifer unit by a genetic algorithm to make data predicted by the model match the actual dynamic production data, and using the corrected volume and water influx of the aquifer unit as aquifer characteristic parameters after the model is automatically fitted; andS300, inverting aquifer distribution: after obtaining optimal solutions of the aquifer characteristic parameters, assigning the volume and water influx of the aquifer unit to a discretized aquifer unit, establishing an aquifer unit bar chart, and determining aquifer distribution by the bar height of each unit.

2. The intelligent identification method of aquifer distribution in complex edge-water oil and gas reservoirs according to claim 1, wherein step S100 comprises: S101, simply characterizing an edge-water oil and gas reservoir: discretizing an edge aquifer into the aquifer unit communicated with the production well unit; discretizing a position with clear geological knowledge into the infill unit; establishing a simple edge-water oil and gas reservoir unit system based on the aquifer unit, the production well unit and the infill unit; and characterizing the aquifer unit by the volume and the water influx of the aquifer unit;S102, based on the aquifer unit, the production well unit and the infill unit, calculating the pressure of each unit by the material balance equation; andS103, according to the calculated pressure of each unit, calculating the saturation and water cut of each unit by using the front advancing equation.

3. The intelligent identification method of aquifer distribution in complex edge-water oil and gas reservoirs according to claim 1, wherein step S200 comprises: S201, establishing characteristic parameter vectors and an objective function of the aquifer unit according to the established numerical simulation model of the water influx unit;S202, setting constraints on the volume of the aquifer unit, and considering that the volume of the aquifer unit is equal to a size of the entire edge-water aquifer; andS203, automatically fitting the volume and the water influx of the aquifer unit by the genetic algorithm, until the error value between the data predicted by the model and the actual dynamic production data is minimum, thereby obtaining the corrected volume and water influx of the aquifer unit.

4. The intelligent identification method of aquifer distribution in complex edge-water oil and gas reservoirs according to claim 1, wherein the material balance equation in S100 is

5. The intelligent identification method of aquifer distribution in complex edge-water oil and gas reservoirs according to claim 1, wherein the waterflood front advancing equation in S100 is

6. The intelligent identification method of aquifer distribution in complex edge-water oil and gas reservoirs according to claim 1, wherein the dynamic production data comprises bottom hole pressure of a single well and water cut of the single well.

7. The intelligent identification method of aquifer distribution in complex edge-water oil and gas reservoirs according to claim 2, wherein the material balance equation in S100 is

8. The intelligent identification method of aquifer distribution in complex edge-water oil and gas reservoirs according to claim 2, wherein the waterflood front advancing equation in S100 is